My 2 cents is this: Why not make a game that ends into another games beginning.  For example, in Civilizations one method of winning is to create a spaceship that will blast off into space.  The game's (Civilization) end game victory condition could then be used to start a game of Galactic Civilizations.  Say you played Civilization to year 2050 or so, you could create multiple colony ships, flagships and/or space miner ships which would give your Gal Civ game a head start versus your opponents.  Also, let your research of technologies in Civilizations carry over to the Gal Civ game instead of just researching the nameless future techs that exist today in Civ.  In other words, start researching the tech tree of Gal Civ while playing the end game of Civilizations.  Further, the Civ world that you create would be your starting world in Gal Civ so any Gal Civ bonuses should reflect the Civilization's game course.

To balance out your material (ships) and technology advantages give the other races some amount of bonuses reflecting their racial backgrounds and the desired difficulty level.

I know that Civ and Gal Civ are made by competing companies and there are other IRL concerns but I was using them as an example for this game succession idea.

To me, there is a natural progression of expansion/research/unit creation that could be tapped for this World Builder followed by a Galaxy Builder concept.

Why do I think that the problem with "War of Magic" was not the magic, but the Heroes. The reason is because Heroes are leveling up. You are basically marry the RPG genre with 4x game. But you forget that a high level hero is basically a superman. When you have a superman and kill the enemy supermen. basically you've win the game. Because the enemy doesn't has anything to counter you anymore. This is the thing that throw the balance off from WOM.

In RPG, you're okay to become a superman. Because the challenge that you have to face will increase with the course of the game. But with 4x game, when you crush the enemy army in a decisive battle, they become weaker. So what happen when you have a superman, and kill the enemy superman? Your enemy basically has lost the game. Because rhey don't have any mean necessary to change their fate.

Plus, a strategy system that use Hero level up system like in WOM will not be a strategy game anymore. You're basically playing an RPG game with a 4x game map. Because you kill the opportunity for the players to use another units. And, just like Heroes Might & Magic Series, you will stick only ro one hero and one party. Other group / army won't be necessary, as it will decrease the opportunity to your main hero to leveling up better (to split the Exp). And once you have already have a superman, then the other unit won't be necessary.

How you ask?

You're now achieving an funnel effect with your armies, they're all nice and together in one small area.

I am in possession of large amount of fire and air shards.

I open with Falling star and it decimates your armies in the formerly mountain hex.

Then follow up with a tornado.

And sent in the cavalry.

If you're like the AI, you have lost your whole invading force in one turn to few cavalry killing your survivors from falling star+ Tornado combo. Ones that survive will be the ones who got flung really faraway by the tornado. It is now your turn!

Plus you're not allowed to alter terraform lands in your enemies anyways. But that is what I would do if you did that.

AoW3 is simply better in all ways. Can't stop playing it.

Here are some possible sources of inspiration that relate to what Frogboy is talking about:

1. AI War: Fleet Command has a textbook example of an effective, believable and challenging "asymmetric-non-AI".  By asymmetric, I mean it's not always trying to play the same game as the player.  By non-AI, I just mean that the AI isn't written to do everything the player needs to do, but it uses some "cheats" in the asymmetry to allow it to not have to.  I'm not doing it justice by the way I'm explaining it, just check it out.  Obviously the theme of the game helps here (your enemy is even call "the AI" in the game), but you can get the same thing in fantasy with magical constructs like undead, familiars, golems, summons, etc, etc.  There's much precedent for fantasy automatons.  Heck, even beasts and magical animals are expected to behave according to "sets of simple rules" like called instincts.

2. Dominions IV is a wonderful example of a TBS that throws balance out the window for the sake of sheer fun and has the most high-powered magic in any 4X fantasy game I've ever played.  Like MOM, it shows that with deliberate game design, it absolutely can be done.  The AI in this game also isn't so great.  As AI goes, I prefer the example above, but at least Dominions is dripping with high-powered magic.

One problem Frogboy didn't mention is that following on the heels of greats like MOM, HOMM, Disciples, etc is that simply being compared to those games holds you to a super-high and super-specific standard.  If I even get the inkling that someone is designing a game "in the spirit of MOM", then implicitly expect it to be at least as satisfying.

Heart Shaped Man's idea is to rather try an approach that's fresh and new enough that one could hopefully avoid such comparisons.  When I read the idea I was a little disappointed because the idea of a modern MOM clone (but hopefully improved in some ways) is extremely enticing.  But that's simply because I've already got this expectation.  Were I to play a well-designed game such that HSM describes, without those expectations, I would probably really enjoy it.

I believe (not done it)  the Paradox Games (Crusader Kings, Europa Universalis, Victoria) have an ability to convert your saves from the end of one to the start of another.

There have been one or two RPGs that did so too.

The AI problem in the Age of the Internet seems well suited for player contributions.

- Either directly by players submitting strategies for inclusion (using a tool to make them relatively easy to convert)

- OR indirectly by harvesting game histories and examining them to see what the players did that the AI can copy.

[esp if you can automate the process  -but that smells perilously of true-AI which probably requires at least three works of outright indisputable genius]

..this would of course require serious after-market support for AI development but since AI doesn't need to be so very hot the first few games and since it's hard to do earlier anyway that might be a good way to plan your AI budget. The AI guy(s) could spend time building the "AI framework" too while the main game 'engine' was being put together so that it was relatively easy to add new strategies to the AI toolbox.  

[I really have no idea what an AI looks like but it seems to me that building such a framework ought not to be impossible and might well be transferable -saleable!- to other games   ...which is good as I suspect it might be a major project in its own right]

For those who don't know, in Dominions: "You are a God! You are master and ruler of a loyal nation. You have unimaginable powers at your disposal. You have claimed this world as yours. But there are others who stand in your way. You must defeat and destroy these pretenders. Only then can you ascend to godhood and become the new Pantokrator." (recently deceased Omnipotent being) 

I haven't played much (didn't like the battle system ..think I played D3though might be worth another look) but I recall needing to have *relatively* mundane garrisons everywhere to keep the worshippers in line and stop people invading while I sent my ridiculously overpowered super-heroes against the enemy forces.

Battles tended to have hordes of soldiers for most of the game and even the Superbeings tended to want support for a long long time (and late-game you don't dare leave them undefended either unless you're really stomping everything with your plethora of -predesigned but extravagantly potent- magic items) Mundane mooks (even the elephants) did become so much chaff eventually and in PVP I can imagine the game going a while in that state.